CatbertTheGreat is back!
Welcome back, Oh Greatest of Sparkly hunters.
Hey, everybody, CatbertTheGreat is here.
The burning question is - are you back for good or just dropping by?
Launch the Rockets!!! Hurray!!!
There's been naught but junk sparklies since he left.
@lavikinga: It's always nice to see @catbertthegreat (just in case he has notification for mentions on). He posted three deals, and the eagle eye of 555 caught the last one.
You guys are funny, this is only temporary.
@shrdlu: If it wasn't for the fact they were rocket deals, I might have thought some one had hacked his account.
@shrdlu: I just got back from lunch or I would have caught the first one. :)
Welcome back CatbertTheGreat!
@catbertthegreat: Don't be a deals tease!
@theoneill555: Lunch? LUNCH? What time zone are you IN, for heaven's sake? It's 1722 here on the west coast.
@wootfast: no he went to play in the desert with rockets.
Now we get can get on with the jewelry postings.
@sgoman5674: Things go very well, thank you. My plants are doing excellent as the California Summer finally begins, and my rockets though shredded to pieces had some glorious last flights. I'm volunteering with a local city council campaign. Summer has gone very well for me so far.
I'll be here till September 23rd and then I'll be off again.
@shrdlu: No I have not done anything with regression analysis, or if I have I don't remember, but most likely not. Still, I wouldn't mind seeing the data though. :)
Thank you everyone for all the lovely words of welcome.
@tygerdave: I think what she's figured out is simple: We're damned if we do. We're damned if we don't.
@catbertthegreat, @shrdlu: Sorry! Had some other junk to attend to. I see that @attillathemom (love that handle) has already provided a link.
The leverage comment should have been a PM , though I bet @hobbit's lady friends from the DAR have some kick-ass hats! So as penance, I'm going to attempt to bring everyone along.
Simply put, linear regression is a statistical technique that fits a line* to a set of points. Why do this? It allows us to describe the effect that changes in something (aka the independent variable) have on something else (aka the dependent variable). What is the effect of payroll size on wins in Major League Baseball? Regression can be used to transform a cloud of data into an answer.
Remember that equation y = mx + b from high school algebra? The "y-intercept" tells you how many wins you'd have if a team played for "the fun of the game" (zero payroll)##, while the slope tells us how many extra wins you get for each million dollars of payroll+++
@catbertthegreat: Well, then. School starts on the 23rd?
Hey, have you done anything with regression analysis? I've been having a good time reverse engineering the reputation (and other leaderboard) systems with it. I have a ton and a half of data, too.
@shrdlu: I hope I'll still be here tomorrow too, I'd rather not fade away in my sleep.
@catbertthegreat: I, too, hope you don't fade away in your sleep. Not yet anyway. You've still got another 60 or 80 years to go. Though, when the time comes, and it almost certainly will, it doesn't seem like such a bad way to go.
Happy thoughts.
Welcome back.
@heymo: My dear, of the lines I hate, I hate that one the most. I forgive you. This time. Next time out, I use the hidden space lasers, and zap you right out of your lazy boy.
@attilathemom: I don't know whether you've noticed, but I have not actually shared any details, although I've mentioned before that I was doing this. I am interested ONLY in the reputation, although sometimes the other leaderboard items have vague interest. Reputation is the only one that lives inside the source for the page (currently mine just dropped from 99.9963684082031 to 99.9891128540039, presumably because I've made a comment elsewhere that has only a single vote--mine). I also have the flounder account for reputation, and am always amused to notice that non-activity is rewarded (for those noticing that inactive people are still on the leaderboard) and activity is punished, so to speak.
No treatises from me, at least not now.
Oooooooooh.... A thread with both @shrdlu and @catbertthegreat! It would have seemed a lovely nostalgic dream just a short time ago. How lovely! I am pleased. Welcome back Catbertthegreat, for as long as you choose to stay.
BTW, I have no idea what @jumbowoot puts into the water that people that leave seem to be inexplicably unable to not resist returning.
@attilathemom: Dang, there are just some seriously FUN people around these parts. Here's some little items, just to entertain you.
I spend a lot of time making comments, and then looking at the source for the page (which gives the raw number for reputation). I go off and vote up newbs, because they're at the other end of the curve from the top 1.5% (which was reported by snapster as the cutoff for who is or isn't black). I don't vote up just anyone, but if there's someone who seems to be making a contribution, AND they aren't being voted up already, then I give them a nudge. Sometimes I go find other comments by that user and vote them up too.
This has the interesting effect of reducing the reputation of all of us in the top rank; how funny is that? It is, after all, a pool, and the more people in the water, the larger, and yet thinner, the distant ends of the curve have to be.
I have about 500 pieces of data related to me and flounder. Fun stuff, yes?
[continued]...
Though regression is a powerful tool, you can get a very wrong answer if your data are funny. Say, for example, you have some values that are very extreme on one or both variables (sometimes called outliers). These outliers can exert undue "leverage" on the line, causing it to change slope (imagine a fat kid on a see-saw). Wrong slope = wrong answer! Good analysis involves lots of diagnostics to make sure your answer is real and not an artifact of a few funny cases or the wrong model.
Whew.
Now, has anyone heard from glinda?
* Or, in the case of multivariate regression, a plane or hyperplane.
## i.e., what value the dependent variable takes on if the independent variable equals zero.
+++ i.e., how much the dependent variable changes for a one unit change in the independent variable.
@heymo: Two things. I point out that I have a well known alter ego, which I work very hard at using only to be kind, or amusing. I prefer that people see the darker, perhaps less pleasant, certainly not always caffeinated me as, well, me. This means that I have two datasets from which to make inferences. In addition, the words regression analysis may be a red herring.
@glindagw is still on walkabout, as far as anyone knows. I believe she said all summer, and summer is only now drawing to a close (for school aged children, in any case, which is who she is on walkabout with). I don't recall whether or not there is home schooling involved, which would imply that the school year is even more fluid.
Perhaps I should drop her an email, offlist, and see if she responds.
what happened? did he forget his password?
@catbertthegreat: Now that you're back, I can leave again for a while. After all, There Can Be Only One.
Hmmmmmmmm
@shrdlu: There is room for both of you!
@catbertthegreat: For how long? And how are things?
I have no idea of Catbert's motivation. Mine was boredom, which is rapidly being overtaken by annoyance. I am ramping back commenting and such right now, and will probably push it back even further once Catbert disappears into the sacred halls of ivy.
So it goes.
[Edit] Not that I'm not fond of the vast majority of people that I interact with, here, but you get the idea.
@catbertthegreat: (and others) I'm quite sure that @heymo was actually referring to a statistical property. As usual, here's a wikipedia link. I was hoping to actually see him show up and defend his statement, but rather than have the rest of you not get to be in on the game, here's further information.
http://en.wikipedia.org/wiki/Hat_matrix
He's correct, of course. Hobbit's values are always skewed. Yours and mine are also, and @lavikinga has entered into the bubble we've discussed before (whether she likes it or not).
@shrdlu: I take lunch when I have a chance and today my chance was from 1540 to 1640 (West Coast).
I also believe that on deals it is not "there can be only one".
I believe that we need at least three to cover all the shifts.
@catbertthegreat: Oh noes! You can't leave again! Even as a newbie I miss you!
@catbertthegreat: To sleep, perchance to dream...
I hope you're still here when I show up again tomorrow morning.
@shrdlu: linear or non linear regression? I actually think a market basket analysis of wooter spending habits could be fun. Or at least funny. What are you looking to prove with this analysis?
sorry, my nerdy analyst ears got all perked up there so I had to horn in.
@attilathemom: Crap, ran out of room (how terribly unusual for me).
What kind of analyst? What math skills?
A brief moment of truth, here. I may or may not actually be using something more similar to a Gaussian distribution, or Bayes, and remind you that this could all just be an elaborate social engineering construct on my part. If you were even vaguely interested, there appear to be AT LEAST three, or perhaps four, items that affect popularity, aka reputation. One of them provides the majority of influence, but the others still contribute, no matter how minor. It becomes interesting at the ends, only (i.e. in the top 3% and the bottom 3%).
Although I believe the reputation algorithm was original written to provide a bell curve, I maintain that the curve is continually skewed, and have provided a link to the proper physical distribution multiple times.
The important thing to take away is that I am using the words RA merely as a placeholder. ;-}
@heymo: I genuinely do not get your second sentence. I already knew your time zone. I could probably come close to the lat/long.
@shrdlu: I don't guess his comment had anything to do with @hobbit's sense of fashion and her fabulous collection of hats, did it? No. I didn't think so. You guess make me think too hard.
@lavikinga: why are we talking about hats?
@shrdlu: I'm guessing he was referring to the black triangle, which can look like a hat.
@shrdlu: ah, not being a regular on these forums I have missed your other discussions of the reputation stats.
I'm actually an operations analyst with a physics background, but a data mining class left me very interested in the properties of data sets, so I like to play with them when I can. Having not bothered to look at any of the data in question I couldn't begin to curve fit though.
@heymo: I second the hate of that phrase (I work for the gummint so I get to hear it a LOT), and I have way better things than lasers to point your way. :P just sayin.
@catbertthegreat: Awesome, you're back!
@heymo: you are thinking of the Red Hat Society. DAR doesn't actually wear hats, except for the old ladies.
Simply put...
YEAH!
@catbertthegreat: Nice to have you back! Any chance for some sparkly geological finds?
@attilathemom: She'd tell you, but then she'd have to kill you.
@catbertthegreat: Welcome back! We all missed you :(
@shrdlu: Sorry. It's the best I could do at 7AM sans caffeine.
Have you run any diagnostics? I'd be concerned about leverage in particular. Bet @hobbit's hat value is downright insane.
[edit] Oh, crap. I just divulged my timezone. Let the triangulation begin.
@shrdlu: I meant it only figuratively: with information sprinkled over the site, you could probably make a decent estimate of my location. But I'll make it easy for you and your space laser of death: 76°44'6.36" W, 39°5'40.90" N
@heymo: Hey, Mo (dammit)! The answer I was hoping for was what you meant by "Have you run any diagnostics? I'd be concerned about leverage in particular. Bet @hobbit's hat value is downright insane."
Cue jeopardy final question music. Doo-doo-doo-doo-doo-doo-doo. Doo-doo-doo-doo-dum-di-doo-doo-doo. Doo-doo-doo-doo-doo-doo-doo. Dum-di-doo-doo. Doo-doo-doo.
Answers. I want explanations of obscure statement, please.
@faughtey: Hah! I KNEW you'd commented in this thread. I finally heard back from DD, and here's the link to the chocolate place in Dallas that I'd mentioned:
http://chocolatesecrets.net/ChocolateSecrets/default.aspx
It wasn't chocolate and jewelry, it was chocolate and wine. Still, it was VERY good chocolate. I just looked at some of their offerings. It's all coming back to me. I'm so very glad I'm three day's drive away.
@attilathemom: You work for the gummint? Contractor, or G-[number]? You don't have to answer, of course, but just curious. Just interesting. Really and truly. You could always send me a PM if you prefer, and I'll provide you with an email. Or not. Up to you.
[Edit] Ugh. It's late, and tomorrow is another long, long day. Off to dreamland for me.
@shrdlu: That looks absolutely decadent. What a great find!
@catbertthegreat: Another big welcome back - I was down and out for a while, not feeling great. Man, did I pick a time to play hooky from Deals-side! I never thought I'd see the day when we'd have another thread with both @shrdlu, and @catbertthegreat!... Glad to hear your updates, and good to see you around again.
Stats
- Asked
- Asked By
- theoneill555
- Answered
- 55 times
- Viewed
- 2535 times
- Voted Against
- 0 times
- Voted For
- 23 times
55 Answers answer
Sort By: